Consistency of EPA Exposure Estimates with Available Monitoring Data

POLLUTION LOCATOR|Consistency of EPA Exposure Estimates with Available Monitoring Data

The best test of the overall reliability of EPA's estimated exposure data would be a nationwide comparison of model estimates with monitored concentrations at the census tract level, for 1996 as well as more recent years. Unfortunately, actual monitoring data exist for only a very small number of HAPs and locations, making an extensive verification of the National-Scale Air Toxics Assessment (NATA) impossible. Currently, monitoring data are available for fewer than 20 HAPs at fewer than 50 locations across the U.S.

EPA PREDICTIONS SYSTEMATICALLY UNDERESTIMATE 1996 MONITORED CONCENTRATIONS
EPA PREDICTIONS ARE SIMILAR TO CURRENT MONITORED CONCENTRATIONS

EPA PREDICTIONS SYSTEMATICALLY UNDERESTIMATE 1996 MONITORED CONCENTRATIONS
To evaluate the reliability of the NATA data, EPA compared modeled concentrations with actual monitoring data for seven HAPs collected by monitoring stations in 1996. The comparison showed that EPA's modeled concentrations consistently underestimate 1996 monitored concentrations. Scorecard also directly compares chemical-specific NATA predictions with historical monitoring data (1995-1997) from stations in the following states:

California (27 monitoring stations, 1996 data)
Maryland (1 station, 1996 data)
Massachusetts (2 stations, 1997 data)
New Jersey (1 station, 1996 data)
New York (10 stations, 1997 data)
Pennsylvania (1 station, 1996 data)
Rhode Island (2 stations, 1997 data)

Possible explanations for NATA's systematic underestimation of historical monitored concentrations include:
1. Gaps in the emissions data used to estimate exposures: emission rates for sources may be underestimated and/or many sources may be missing from emissions inventories.
2. Bias in the siting of monitoring stations: many monitoring stations have been sited to find peak concentrations.

EPA PREDICTIONS ARE SIMILAR TO CURRENT MONITORED CONCENTRATIONS
Comparisons of EPA model estimates with the limited monitoring data available indicate that exposure estimates based on 1996 emissions data are a relatively good indicator of current air quality conditions at the county scale. Scorecard directly compares chemical-specific NATA predictions with current monitoring data (2000-present) from stations in the following states:

California (17 monitoring stations, 2001 data)
Colorado (7 monitoring stations, 2001 data)
Connecticut (4 monitoring stations, 2001 data)
District of Columbia (2 monitoring stations, 2001 data)
Florida (6 monitoring stations, 2001 data)
Georgia (4 monitoring stations, 2001 data)
Idaho (1 monitoring station, 2001 data)
Illinois (13 stations, 2002 data)
Indiana (3 stations, 2002 data)
Louisiana (8 monitoring stations, 2001 data)
Maryland (8 stations, 2001 data)
Massachusetts (2 stations, 2001 data)
Maine (2 monitoring stations, 2001 data)
Michigan (8 monitoring stations, 2001 data)
Minnesota (5 stations, 2000 data)
Missouri (1 monitoring station, 2001 data)
Montana (2 monitoring stations, 2001 data)
Nebraska (2 monitoring stations, 2001 data)
New Jersey (4 stations, 2002 data)
Ohio (13 stations, 2001 data)
Pennsylvania (3 stations, 2000-2002 data)
South Carolina (28 monitoring stations, 2001 data)
South Dakota (1 station, 2001 data)
Tennessee (9 monitoring stations, 2001 data)
Texas (58 monitoring stations, 2001 data)
Utah (1 monitoring station, 2001 data)